Jiaqiang Li

Output—

Output-volume data is insufficient for scoring: the OpenAlex record lists only 3 works in the recent three-year window, too thin to characterize productivity. This axis would measure publication volume, venue tier, and share of first- or corresponding-author roles.

Impact—

Citation-impact data is not yet sufficient for scoring: the record carries h-index 1 and i10-index 0, too early to reveal any citation pattern. This axis would measure total citations, citing works, and downstream influence of the two reviewed papers.

Focus0/100

Focus scores 0/100 because the field distribution is an even split: 50% multimodal AI evaluation on scientific-reasoning benchmarks and 50% AI agents and human-AI collaboration. Topic counts echo the breadth — Topic Modeling (3), Natural Language Processing Techniques (2), Multimodal Machine Learning Applications (2), plus single appearances of neural-network applications and multi-agent negotiation. At this career stage the two strands read as complementary evaluation work rather than unfocused drift.

Collaboration87/100

Scoring 87/100, collaboration is this profile's strongest dimension. The OpenAlex record links Jiaqiang Li to 78 distinct coauthors, and twelve recur frequently — Yajie Yang, Zhiheng Xi, Jiadong Chen, Enyu Zhou, Senjie Jin, and others — indicating a stable core team within a wide peripheral network. Both catalog papers are team efforts: first-author on Sci-MMR and last-author on Atria Dawn, showing Li operates across authorship positions inside large groups.

Activity40/100

Activity scores 40/100. OpenAlex counts-by-year shows 2 works in 2025 and 1 in 2026 — three works across the recent three-year window — a modest but continuous cadence for an early-stage researcher. Both catalog papers are 2026 arXiv preprints produced with large teams, so current output is concentrated in collaborative bursts rather than a steady solo stream; the small base makes the 40 score provisional.

Reproducibility38/100

Reproducibility scores 38/100, averaged over the 2 papers Evolve has reviewed (reproMean 38, reproPapers 2). Sci-MMR rates medium readiness — unusually well-audited but constrained by 235 tasks, LLM-judged coverage and unquantified variance — while Atria Dawn rates low, with its benchmark table and self-reported attribution numbers flagged for cherry-picking risk. The split suggests careful evaluation practice on benchmark work, not yet consistent artifact transparency across outputs.